On solving the planar \(k\)-centrum problem with Euclidean distances (Q613427)

The authors deal with the \(k\)-centre problem in the plane. This continuous location problem belongs to the family of Euclidean convex ordered median problems, where such point in the plane is searched, which minimizes an objective function depending not only on weights and distances from customers to the searched point, but even on an order of these distances. The objective function of the \(k\)-centre problem is defined as a sum of the k largest distances. The authors based their approach on a gradient descent method similar to Wieszfeld's iterative method. They complied with the fact that the objective function is pointwise defined by introducing bisector curves, which separate the part of plane, where the order of the distances from fixed customers to points does not change. They treated also the particular case of equal weights associated with the demand points. Finally they adapted the studied procedures for solving the \(k\)-centre problem. The concluding computational analysis shows that their proposed procedures provide better results than the existing ones, as concerns the number of iteration and the computational time.